Actin architecture in moving nanofibers

We coated thin glass nanofibers with GST-tagged WCA domains (GST-WCA) from N-WASP and observed formation of individual actin filaments with TIRF microscopy in the presence of profilin-actin, Arp2/3, and CP (Figure S1). We excluded cofilin, VASP, and actin bundling proteins from these experiments to understand how a minimal set of proteins direct filament organization in moving nanofibers. Although cofilin is an important player in dendritic nucleation, cofilin acts downstream of the dendritic nucleation pathway. For in vitro motility assays, cofilin increases propulsion rates in long-term studies by maintaining an actin monomer pool [5], [50]. We therefore restricted our initial observations to the first 30 to 90 minutes of comet tail formation. Experiments with and without cofilin showed little difference in particle velocities in similar motility buffers for these short-term observations (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Average particle velocities in 8.5 µM profilin-actin, 100 nM Arp2/3.

https://doi.org/10.1371/journal.pone.0031385.t001

In well-blocked flow-cells, nanofibers rarely adhered to the top surface but fell to the chamber bottom. We thus required a TIRF microscope with evanescent excitation at the chamber bottom to observe moving nanofibers. In contrast, previous prism-based TIRF microscopy observations of actin comet tails generated NPF-coated glass nanofibers [51] were restricted to nanofibers permanently affixed to the chamber top. Consequently, these experiments were limited to observations of actin architecture during branch initiation and not during sustained nanofiber motility.

Actin shells formed around nanofibers in motility buffer at actin concentrations of 8.5 µM and a CP to Arp2/3 ratio of 0.75 to 2, often before the nanofiber drifted to the chamber bottom and into the TIRF microscopy excitation field. After initial shell formation, the nanofibers usually broke symmetry to form either one or two comet tails. Though 68% of observed nanofiber initially formed one comet tail of polymerizing actin, 32% initially formed two comet tails, one on either side of the nanofiber axis (Figure 1F, H). In the latter case, one tail usually attained dominance and the residual tail was left behind near the original shell as the nanofiber moved (Figure 1B). In rare cases, two tails persisted on either side of the nanofiber, and the nanofiber remained relatively stationary while the tails were pushed away from the nanofiber by new filament growth at the nanofiber surface (Figure 1G). For both single and double comet tails, the majority of new filament growth appeared at the nanofiber surface, while the network within the comet tail either remained stationary or was pushed rearward at a slow rate. Thus, the majority of barbed end growth was directed toward rather than away from the nanofiber surface, consistent with previous reports [21], [46], [50] and the dendritic nucleation model.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Actin filament and branch geometry in comet tails under TIRF microscopy.

Conditions: A–B, 8.5 µM (8% labeled) actin, 9 µM profilin, 100 nM Arp2/3, CP as indicated; C–E, 2 µM (10% labeled) actin, 3 µM profilin, 20 nM Arp2/3, 40 nM CP; F–H, 8.5 µM (8% labeled) actin, 9 µM profilin, 100 nM Arp2/3, CP as indicated; nanofibers coated with 10 µM GST-WCA from N-WASP, motility buffer, 0.38 mM total ATP. (A–B) Actin architecture in comet tails (T) of moving nanofibers (dashed outline) was visible under TIRF microscopy. In 100 nM CP, comet tails consisted primarily of long filament bundles. Increasing CP to 200 nM generated a branched actin networks with short bundles (Black arrowhead). (C) Lowering profilin-actin, Arp2/3, and CP concentrations showed individual filaments and branches (white arrowheads) in the comet tail (T). Some filaments (white arrows) crossed the nanofiber boundary, while others terminated at the nanofiber (black arrows). Brighter filament bundles (black arrowhead) terminated at the nanofiber. (D) Epi-fluorescence image of panel C. (E) Magnified image of box in B showing bundle (black arrowhead) dissociation. The bundle was formed from daughter filaments from the same mother filament (white arrowheads). (F) In high CP, nanofibers sometimes formed two comet tails. (G) Kymograph of line in F showing tail expansion at the nanofiber surface (dashed outline) under TIRF (left) and DIC (right) microscopy. (H) Nanofibers sometimes formed two comet tails in low CP. Long actin bundles (black arrowheads) appeared within and beyond the comet tails. Scale bars are 1 µm for E and 5 µm for all others. Times are shown in min∶sec.

https://doi.org/10.1371/journal.pone.0031385.g001

Nanofibers with single tails moved at 0.14 µm/min at an optimal CP to Arp2/3 ratio of 1.0 and slightly slower at a CP to Arp2/3 ratio of 2.0 (Table 1). These speeds were substantially slower than previous reports of 2.2 µm/min movement of spherical particles in similar buffers lacking cofilin [50]. Much of this difference could be attributed to our use of muscle actin, which produced 4-fold slower particle speeds than did cytoplasmic actin [50]. In support, we found that 4.5 µm diameter spherical particles moved at 0.48 µm/min in the same motility buffers used for nanofibers (Table 1). These rates were comparable to previous studies in which similarly sized beads moved at 0.5 µm/min in similar motility buffers containing muscle actin [52]. Thus, the 4-fold difference between nanofiber and bead speeds in our motility buffers was due to particle geometry, and likely attributable to the thin but wide comet tails generated by 200 nm thick nanofibers.

Capping Protein controls the transition between bundled or branched actin networks

Previous studies showed that actin filaments can form from dendritically nucleated networks in vitro and that CP can control the proportion of branched to bundled filaments [53], [54]. However, filament barbed ends in these studies grew away from the particle surface and bundling did not result in particle motility. We found that comet tails with barbed ends directed toward WCA-coated nanofibers also contained two populations of actin filaments. Under TIRF microscopy, bundled filaments resembling microspikes primarily formed perpendicular to the direction of movement within a dendritic network of highly branched filaments. In contrast, these bundles were invisible under epi-fluorescence microscopy (Figure 1D). The variance in actin network architecture between bundled and branched filaments persisted as comet tails grew at the nanofiber interface. The dominance of these two structures in comet tails changed with different concentration of CP. Actin filament bundles were more prominent at lower CP concentration (100 nM) (Figure 1A) while branched filaments dominated at higher CP concentration (200 nM) (Figure 1B).

Actin branches and bundles were inter-convertible. Although comet tail formation was rare at lower actin concentrations of 2 µM, the origins of bundles and branches were more easily discerned in the few, sparse comet tails that formed. Here, individual filaments converged to form brighter bundles (Figures 1E left panel) and bundles could sometimes be seen to dissociate into individual daughter filaments emerging from the same mother filament (Figure 1E right panel).

Bundles terminate at the nanofiber surface

At low (≤75 nM) CP concentrations, actin filament bundles frequently projected beyond the nanofiber surface as they elongated (Figure S5). In higher CP (≥100 nM), the majority of bundles continuously terminated at the nanofiber as they grew (Figure 1A–C black arrowheads). In contrast to bundles, individual actin filaments within the comet tail often elongated across and beyond the nanofiber surface, even at higher CP concentrations (Figure 1A, C arrows).

Bundles elongated faster than the surrounding comet tail, consistent with previous studies in which filament bundling proteins increased motility speeds [5], [28], [47]. For example, four bundles from Figures 2A–B elongated at 0.22±0.05 µm/min, while the overall comet tails expanded at 0.09±0.01 µm/min. Faster bundle elongation often resulted in bending or buckling of the bundle between the fast growing barbed ends at the nanofiber surface and the pointed ends embedded within the dendritic network (Figure 2A, B arrowheads).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Filament bundles processively attach to the nanofiber.

Conditions: 8.5 µM (8 to 20% labeled) Mg-ATP actin, 9 µM profilin, 100 nM Arp2/3, and indicated CP. Nanofiber coating and buffers as in Figure 1. (A–B) Bright filament spots grew against the nanofiber surface. Spots either remained attached to the same location on the moving nanofiber (A, black arrowhead) or oscillated back and forth along the moving nanofiber (B, blacks arrowhead). Bright spots grew faster than the surrounding comet tail network (T) and often buckled from compression between the nanofiber and tail network (arrow). White arrowhead in B indicates bundle starting position. (C) CP was lowered from 200 to 50 nM after the establishment of two comet tails (T) by motility buffer exchange. Short, bright bundles (top panels, black arrowheads) became rapidly polymerizing bundled loops (black arrowheads, bottom panels) that remained attached to the nanofiber at their growing ends (white arrowhead) and to the tail at their pointed ends (white arrows). Growing bundles splayed into filaments of different loop lengths. (D) Traces of bundles lengths over time before CP reduction. Zero seconds represents the point of buffer exchange, 163 min after the experiment start. (E–H) Traces of individual filament lengths over time after CP reduction showed that filament within each bundle grew at different rates. Plots represent filaments from the same bundle. Some filaments continuously elongated while others show pulsed growth. Scale bars are 1 µm for A–B and 5 µm for C. Times are shown in min∶sec for A–B and hr∶min∶sec for C.

https://doi.org/10.1371/journal.pone.0031385.g002

Processive or semi-processive attachment between growing bundled barbed ends and the nanofiber surface was further illustrated by the reduction of CP concentration through an exchange of motility buffer on an existing comet tail (Figure 2C). Here, short bundles present before the buffer exchange (top panels) elongated substantially after CP concentration was reduced (bottom panels). Processive attachment of bundled barbed ends to the nanofiber surface coupled with their faster growth rates (2.4 µm/min) resulted in significant looping of these bundled filaments. We observed similar buckling and looping of bundles on moving nanofibers (Figure S5, Table 2). Thus, bundled barbed ends were not physically trapped by strong attachments between the nanofiber and the chamber surface, but were likely attached through specific molecular interactions with GST-WCA at the nanofiber surface.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Frequency of bundle formation on nanofibers.

https://doi.org/10.1371/journal.pone.0031385.t002

As bundled loops grew, they split into individual filaments of varied lengths. Measurement of individual filament growth rates from several bundles showed that filaments within a bundle grew in a salutatory fashion (Figure 2E–H), indicating that capping times varied between individual barbed ends within the bundle. During periods of elongation, bundled filaments grew 10-fold faster (2.4±1.1 µm/min, N = 7) after CP reduction than they did before CP reduction (0.202±0.002 µm/min, N = 2, Figure 2D). While some filaments within a bundle elongated, others from the same bundle periodically halted. This lack of barbed end growth was likely due to capping either by CP or by WH2 domain attachment to the hydrophobic cleft of one of the barbed end subunits. In the latter case, WCA domain attachments could maintain the connection between the bundle and the nanofiber.

Cellular levels of magnesium generate actin bundles in vitro

Actin bundles or microspikes form at the leading edge of motile cells in vivo [55] and bead-based in vitro motility experiments have shown that actin bundles generate more propulsive force than individual branched filaments [5], [28], [47]. Our in vitro experiments indicated that actin filament bundles were prevalent even in the absence of specific actin crosslinking or bundling proteins in our motility buffers. We therefore sought to eliminate bundles to determine the relative contributions of dendritic and bundled filaments to particle motility.

Both the methylcellulose used in most in vitro TIRF microscopy assays of actin dynamics [56], [57] and polyvalent cations such as Mg²⁺ [58] can bundle filaments. However, the mechanism of bundle formation by these two solutes differs substantially [59]. Methylcellulose contributes to bundle formation through entropic depletion forces that reduce effective solution volume. Polycations overcome the 14 e⁻/subunit negative charge of filamentous actin to promote filament side-to-side association.

One study of the concentration dependence of methylcellulose on bundling [60] found that filaments start to bundle in high viscosity methylcellulose (88 kDa, 4000 cPs at 2%) at a threshold concentration of 0.2%. Similar studies with polyethylene glycol showed that the filament bundling threshold also depends upon the chain length of the excluding solute [61], [62]. Although we used 0.25% methylcellulose in our TIRF microscopic assays to stabilize actin filaments for imaging, we used a lower viscosity species (63 kDa, 1500 cPs at 2%). Thus, the contribution of methylcellulose to bundling was unclear but likely near the threshold concentration.

Actin paracrystals formation requires Mg²⁺ concentrations of ≥10 mM [58], but several studies have reported anecdotal evidence that bundles form at the 1 mM Mg²⁺ concentrations used in typical TIRF actin microscopy and in vitro motility buffers once filament densities increase [49], [53], [63]. We therefore sought Mg²⁺ concentrations that would eliminate filament bundling in the presence of minimal, 0.25% of 63 kDa methylcellulose.

In typical actin polymerization assays, the Ca²⁺ bound to the high affinity Mg²⁺ site on actin is first exchanged with Mg²⁺ by addition of an excess of Mg-EGTA prior to addition of KCl. To generate low Mg²⁺ conditions, we diluted Ca-ATP-actin monomers into buffer without added Ca²⁺ and reduced Mg-EGTA concentrations 5-fold during the exchange step. We correspondingly reduced ATP concentrations from 0.38 mM used above to 0.2 mM, a concentration sufficient for both actin polymerization and Arp2/3 nucleation. Total Mg²⁺ concentration in the assay was regulated by subsequent addition of MgCl₂. We calculated free Mg²⁺ concentration from the pH and total buffer CaCl₂, EGTA, MgCl₂, ATP, and KCl concentrations using existing methods [64], [65]. At standard 1 mM total Mg²⁺ concentration, filament bundles could be seen after 30 min of polymerization (Figure 3A). An increase in total Mg²⁺ to either 5 or 10 mM increased both the speed of bundle formation and thickness of bundles. Reduction of total Mg²⁺ to 0.5 mM (0.3 mM free) or below severely reduced bundle formation in the presence of 0.25% methylcellulose. Quantification of bundle formation under a range of Mg²⁺ concentrations (Figure 3B) showed that bundles, assessed as overlap of two or more filaments, rarely formed at or below 0.5 mM Mg²⁺ but formed readily at concentration of 1 mM and above. Correspondingly, bundle onset time (Figure 3C) was significantly shortened at high Mg²⁺ concentrations. As the typical cellular Mg²⁺ ranges from 0.5 to 1.5 mM [66], [67], [68], we expect that similar Mg-actin bundles form at the leading edge where filament density is high.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Cellular Mg²⁺ concentrations bundle actin filaments at high densities.

Conditions: 2.5 µM (20% labeled) Mg-ATP actin, 50 mM KCl, 1.05 mM EGTA, 10 mM imidazole, pH 7.0, 100 mM DTT, 0.2 mM ATP, 15 mM glucose, 20 µg/ml catalase, 100 ug/ml glucose oxidase 0.25% 1500 cP methylcellulose. (A) Time-lapse TIRF microscopy images of de novo nucleated actin filaments. Images in each column were taken at the same time (min∶sec) after addition of salts. MgCl₂ was added to low Mg-EGTA polymerization buffer to set the total Mg²⁺ as indicated. Free [Mg²⁺] was calculated from pH, ionic strength, and total Ca²⁺, Mg²⁺, EGTA, and ATP. Actin bundles readily formed at 1 mM total Mg²⁺ once filament densities increased. Increasing total Mg²⁺ to 5 or 10 mM increased the speed and extent of bundle formation. (B) Fraction of filaments forming bundles over time. Free Mg²⁺ concentration of at least 0.7 mM significantly increased bundle formations. (C) Time at which the first bundle was observed as a function of Mg²⁺ concentration. Scale bar is 2 µm.

https://doi.org/10.1371/journal.pone.0031385.g003

Reduction of Mg²⁺ below cellular levels abolishes motility

Remarkably, lowering Mg²⁺ 10-fold to 0.1 mM total (0.03 mM free) completely abolished nanofiber motility (not shown). Restoration of Mg²⁺ to 1 mM total (0.7 mM free) with additional MgCl₂ restored nanofiber motility. While glass nanofibers provided a good model of a leading edge, they varied in length, diameter, and curvature. Thus, analysis of nanofiber motility speeds as a function of Mg²⁺ concentration could be substantially influenced by variations in nanofiber geometry. As an alternative to nanofibers, we used 4.5 µm diameter polystyrene microspheres to quantify the dependence of particle motility on Mg²⁺ concentration. We found that bead motility speeds also varied greatly with free Mg²⁺ concentration (Figure 4A, C). Reduction of total Mg²⁺ concentration to 0.5 mM (0.3 mM free) or below abolished comet tail growth from GST-WCA coated beads.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Polycations or fascin are required for bead motility.

Conditions: 8.5 µM (20% labeled) Mg-ATP actin, 9 µM profilin, 100 nM Arp2/3, 200 nM CP, 4.5 µm diameter bead coated with 8.5 µM GST-WCA, low Mg²⁺ buffer (50 mM KCl, 0.105 mM MgCl₂, 1.05 mM EGTA, 10 mM imidazole pH 7.0, 100 mM DTT, 0.2 mM ATP, 15 mM glucose, 0.25% methylcellulose, 20 µg/ml catalase, 100 ug/ml glucose oxidase) supplemented with MgCl₂, Lys-Lys·2HCl, KCl, or fascin as indicated. (A) TIRF and Epi-fluorescence microscopy images show representative actin comet tails (T) grown from GST-WCA coated beads (B, dashed circle). All images were recorded 40 minutes after the reaction start. In low Mg²⁺, beads formed a shell (S) that broke symmetry but rarely a comet tail. Tails that did form remained short and detached from the bead. Restoration of cellular, 1 mM Mg²⁺ restored comet tail growth. Additional Mg²⁺ accelerated comet tail growth. (B) TIRF microscopy images of actin comet tails grown in 0.1 mM total Mg²⁺ with added Lys-Lys²⁺ as indicated. All images were recorded 40 minutes after the reaction start. Lys-Lys²⁺ substituted for Mg²⁺ to restore motility. (C–D) Comet tail growth over time after the reactions start. The lengths of actin comet tails from A–B were recorded in each frame. Line segments represent growth of individual comet tails. Comet tail growth increased with the concentration of divalent cation, either in the form of (C) Mg²⁺ or (D) Lys-Lys²⁺. (E) Comet tail growth rates from C–D as a function of free cation. Both MgCl₂ and Lys-Lys-2HCl restored motility in a concentration dependent manner. Removal of methylcellulose did not influence the trend of comet tail growth rates as a function of Mg²⁺. Addition of 5, 10, or 15 mM KCl did not restore motility in low Mg²⁺ buffers. (F–G) Comet tail growth over time in low Mg²⁺ with added fascin. Line segments represent individual comet tails. (F) In low, 0.03 mM free Mg²⁺, 80 nM fascin optimally restored motility while (G) 500 nM fascin optimally restored motility in 0.3 mM free Mg²⁺. Line breaks (arrows) in no fascin represent growth of an actin shell followed by shell detachment during an observation. (H) Comet tail growth rates from F–G as a function of fascin concentration. Errors bars in E and H show S.D. of tail growth rates. Scale bars in A–B are 5 µm.

https://doi.org/10.1371/journal.pone.0031385.g004

Restoration of total Mg²⁺ to cellular concentration level of 1 mM total (0.7 mM free) restored comet tail growth rates to 0.16±0.07 µm/min (Figure 4E). Additional of Mg²⁺ above 1 mM accelerated comet tail growth rates in a concentration-depended manner. For example, 10 mM total (9.4 mM free) Mg²⁺ accelerated comet tail growth rates to 0.59±0.17 µm/min, an approximate 4-fold increase over the rate in 1 mM Mg²⁺. Comet tail growth showed a similar dependence on Mg²⁺ in the absence of methylcellulose (Figure 4E). These results indicate that cellular levels of Mg²⁺ are necessary for generating in vitro motility and that entropic depletion forces play little to no role in this requirement.

Actin binding proteins show little Mg²⁺ dependence

Actin polymerization in KCl has been shown to be independent of excess magnesium [69]. We sought to test the possibility that Mg²⁺ was necessary for some aspect of in vitro motility other than actin filament bundling. We thus measured actin polymerization, Arp2/3 nucleation, and CP binding using bulk pyrene actin assembly assays in the same range of Mg²⁺ concentrations as our TIRF microscopy assays of bead motility (figure S4). As expected, varying free Mg²⁺ concentrations over a 300-fold range had no effect on KCl induced actin polymerization or Arp2/3 nucleation. Polymerization of profilin-actin from capped filament seeds showed a slight Mg²⁺ dependent increase in apparent final filament concentration. However, the initial slope of polymerization, an indicator of concentration of free ends, decreased slightly with Mg²⁺ concentration, and CP was the least active in low Mg²⁺ concentrations that abolished bead motility. Thus, abolishment of motility in low Mg²⁺ and restoration in high Mg²⁺ was not due to the effect of Mg²⁺ on individual components of the motility machinery.

Cofilin does not rescue polycation-dependent motility

At high Mg²⁺ concentrations, comet tail elongation slowed over the course of each 140 min experiment (Figure 4C–D). This decrease was likely due to a decrease in available ATP-actin monomers over time from their incorporation into filaments. To test whether this decrease in monomers influenced Mg-dependent comet tail formation, we added cofilin to the motility buffer to establish a steady state of free actin monomers (Figure 5). Cofilin addition abolished the gradual decrease in comet tail elongation over time seen at high Mg²⁺ concentrations but did not restore motility in low Mg²⁺ buffers. As with motility in buffers lacking cofilin, addition of 0.7 mM or more free Mg²⁺ restored motility in the presence of cofilin. Average comet tail growth rates were slightly higher in cofilin than without, but growth rates showed approximately the same dependence on free Mg²⁺ with or without cofilin (Figure 5C). Coupled with the formation of both primary and secondary actin shells around beads in low Mg²⁺ buffers (see below), the lack of comet tail formation in low Mg²⁺ buffers without cofilin was not due to a lack of polymerization competent actin monomers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Polycation-dependent motility does not require cofilin.

Conditions: 8.5 µM (20% labeled) Mg-ATP actin, 9 µM profilin, 100 nM Arp2/3, 200 nM CP, 2 µM cofilin, 4.5 µm diameter bead coated with 8.5 µM GST-WCA, low Mg²⁺ buffer as in Figure 4 supplemented with MgCl₂ as indicated. (A–B) Comet tail growth over time after the reactions start in the presence (A) and absence (B) of cofilin. The lengths of actin comet tails were recorded in each frame. Line segments represent growth of individual comet tails. Comet tail growth increased with the concentration of Mg²⁺, either in the presence or absence of cofilin. (C) Comet tail growth rates from A–B as a function of free Mg²⁺. MgCl₂ restored motility in a concentration-dependent manner in the presence or absence of cofilin. Errors bars show S.D. of tail growth rates.

https://doi.org/10.1371/journal.pone.0031385.g005

Di-lysine restores bead motility in sub-cellular concentration Mg²⁺ buffers

Based on the above observations, we reasoned that if in vitro bead motility depends upon the actin bundling activity of Mg²⁺ and not its coordination of ATP hydrolysis, then other polycations that bundle filaments should restore bead motility in low Mg²⁺ buffers. Oligomers of lysine have been shown by light scattering to bundle actin filaments [58], with the extent of bundling dependent upon the number of lysine residues. To match the effects of Mg²⁺ on motility as closely as possible, we added millimolar concentrations of the divalent cation Lys-Lys²⁺ to low (0.03 mM free) Mg²⁺ motility buffers. Because lys-lys²⁺ is too large to fit into the high-affinity ATP binding cleft, it would not affect Mg-ATPase activity of either Arp2/3 or actin.

Addition of 0.5 mM Lys-Lys²⁺ to low Mg²⁺ buffers did not restore bead motility, but addition of 1 to 10 mM Lys-Lys²⁺ restored bead motility in a concentration-depended manner (Figure 4B, Figure S2). Although Lys-Lys²⁺ did not restore bead motility to the same extent as additional Mg²⁺ (Figure 4D–E), Lys-Lys²⁺ is a zwitterion at cellular pH and not a true divalent cation. We expect that the negative charge of the deprotonated carboxyl group of Lys-Lys²⁺ limited the extent of its bundling activity. In contrast, addition of 5, 10, or 15 mM of monovalent KCl did not restore bead motility in low Mg²⁺ buffers (Figure 4E). Thus, restoration of bead motility by Lys-Lys²⁺ was not simply due to an increase in buffer osmolarity or ionic strength, and in vitro motility absolutely requires polyvalent cations such as physiological concentrations of Mg²⁺.

Filament bundling by fascin restores processive motility

To test if the dependence of in vitro motility polycations was due to actin bundling, we added the bundling protein fascin to low Mg²⁺ motility buffers and measured comet tail lengths generated by GST-WCA coated beads. In very low, 0.03 mM free Mg²⁺, addition of 80 nM fascin optimally increased comet tail growth rates (Figure 4F, H, figure S3). Addition of 500 or 1000 nM fascin restored comet tail growth but to a lesser extent than did 80 nM fascin.

TIRF microscopic images of comet tails with added fascin (figure S3) showed a subset of straight, bright filament bundles within the comet tail that persisted throughout the entire observation period. These rigid fascin bundles were consistent with fully coupled bending in which filaments are rigidly adhered by specific crosslinks [70]. In contrast, magnesium bundles were highly dynamic and curved (Figure 2), consistent with decoupled bundle bending and non-specific polycation-mediated adhesion between actin filaments. Thus, the failure of fascin to fully restore motility could be due to differences between fascin and magnesium bundle rigidity. In support, addition of fascin in vitro has been previously shown to slow initial particle motility [71], although this effect could be due to a reduction Arp2/3 mediated branching [72].

To address different contributions of fascin- and magnesium-bundled filaments to motility, we increased the free Mg²⁺ to 0.3 mM, a concentration that did not support motility on its own (Figure 4G–H, figure S3). Addition of 80 nM fascin restored motility to the same extent as in 0.03 mM free Mg²⁺, but addition of 500 or 1000 nM fascin restored motility to a greater extent than in the lower Mg²⁺ concentration. Although fascin did not restore motility to the same extent as did high Mg²⁺, the difference in stiffness and persistence between fascin and magnesium bundles, coupled with competition between fascin and Mg²⁺ for inter-filament binding could explain the difference in comet tail elongation seen between fascin and high Mg²⁺.

Filament bundling mediates sustained comet tail attachment

Beads in low Mg²⁺ grew thick actin shells that eventually broke symmetry (Figure 4A, bottom panel). However, the resulting nascent comet tails did not elongate but completely detached from the bead surface and floated away (Figure 6A). Comet tail detachment was not due to a lack of polymerization competent actin or Arp2/3, as a secondary actin shell often formed on the bead surface after the primary shell detached (Figure 6B–C). We interpreted this comet tail detachment as a failure to establish rapid rebinding of barbed ends to WCA domain in moving particles in the absence filament bundling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Divalent cations or fascin rescues comet tail attachment.

Conditions as in Figure 4. (A) Time-lapse epi-fluorescence and TIRF microscopy sequence showing detachment of primary actin shell (1°S) from GST-WCA coated bead in 0.03 free Mg²⁺. Filament density between the shell and the bead surface (dotted circle) is gradually lost. (B) Epi-fluorescence fluorescence microscopy showing formation of secondary actin shell (2°S) after detachment of primary actin shell (1°S) in low Mg²⁺. Times are shown as min∶sec. Scale bars are 5 µm. (C) Kymograph of line in B showing the detachment of primary shell (1) and establishment of a secondary shell (2). (D–O) Percentage of GST-WCA coated beads with either an actin shell or comet tail over time. At the reaction start, all beads developed a thin actin shell. In low 0.03 mM free Mg²⁺ buffer, actin shells detached over time (D) and many beads formed a secondary actin shell (E). Addition of 0.3 mM (F), 0.7 mM (G), 4.5 mM (H), or 9.4 mM (I) free Mg²⁺ restored shell or comet tail attachment. Addition of either 1 mM (J), 5 mM (K), or 10 mM (L) Lys-Lys²⁺ restored actin shell or tail attachment in 0.03 mM free Mg²⁺ buffer. Addition of 80 nM (M), 0.5 µM (N), or 1 µM (O) fascin restored actin shell or tail attachment in 0.03 mM free Mg²⁺ buffer. Means and S.D. were calculated from three independent experiments. At least 50 beads were counted in each experiment.

https://doi.org/10.1371/journal.pone.0031385.g006

To test the dependence of attachment on bundling, we quantified the fraction of GST-WCA coated beads with attached actin shells or comet tails over time (Figure 6D). Of the beads that initially formed actin shells in 0.03 mM Mg²⁺ buffers, these shells were eventually lost so that by 90 minutes only 15% of beads retained their primary shell and 55% of beads had formed secondary shells (Figure 6E). Additional of Mg²⁺ restored bead-actin attachment in a concentration dependent manner (Figure 6F–I). Increased comet tail growth rate was correlated with increased comet tail attachment. Therefore, Mg²⁺ induced bundling was likely required to maintain comet tail attachment to WCA at the bead surface to sustain bead motility. Similarly, addition of lys-lys²⁺ (Figure 6J–L) or fascin (Figure 6M–O) also restored sustained attachment of comet tails to the bead surface in 0.03 mM free Mg²⁺ buffers in a concentration dependent manner. Restoration of attachment was most dramatic in 1 µM fascin, as 100% of beads retained their comet tails over 90 minutes of observation. Thus, filament bundling did not mediate actin shell formation but continued attachment between actin filaments and the bead surface once motility was established.

Bundling enhances barbed end binding to WCA domains in the absence of Arp2/3 and CP

To better dissect the role of filament bundling played in barbed ends binding to leading edge, we captured pre-formed filaments by GST-WCA coated beads in the absence of profilin, Arp2/3, and CP. Fluorescent actin was polymerized in low Mg²⁺ buffer and sheared to increase the number of ends. Short filaments were then incubated for 10 minutes with GST-WCA coated beads, actin monomers in low Mg²⁺ buffer, and added Mg²⁺, Lys-Lys²⁺, or fascin. After the beads were separated by centrifugation, filaments captured along the bottom surface of the bead were imaged by TIRF microscopy (Figure 7A). We scored filaments or bundles that terminated at the bead surface as captured barbed ends (Figure S6). Although we could readily discern individual filaments from bundles under TIRF microscopy, we could not accurately determine the number of filaments within each bundle. We therefore scored bundles as having two barbed ends for the purpose of estimating the total number of captured ends. Counts of total filament ends captured per bead (Figure 7B) showed that additional Mg²⁺, Lys-Lys²⁺, or fascin increased barbed end capture by tethered GST-WCA. Barbed end capture in low Mg²⁺ was only slightly higher than the number of filaments that coincidentally overlapped with control bovine serum albumin (BSA) coated beads (0.9±0.2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Bundling promotes barbed end attachment to WCA domains in the absence of Arp2/3 complex.

Conditions: 8 µM (30% labeled) actin was polymerized in low Mg²⁺ F buffer. Filaments were incubated with GST-WCA coated microspheres with 1 µM ATP-actin monomers and the indicated final concentration of Mg²⁺, Lys-Lys²⁺, fascin, or K⁺ for 10 minutes. Beads were centrifuged, resuspended in low Mg²⁺ buffer in the absence of methylcellulose, and imaged on poly-lysine coated coverslips. (A) TIRF microscopy images of actin filaments and bundles attached to the bottom of coated microspheres in the indicated concentration of Mg²⁺, Lys-Lys²⁺, fascin, or K⁺. The number of actin filaments and bundles crossing the bead boundary (dashed circle) were counted for each bead. Scale bar is 2 µm. An example measurement is shown in Figure S6. (B) Count of average number of captured filaments per bead as a function of Mg²⁺ (○), Lys-Lys²⁺ (•), fascin (□), or K⁺ (⋄) concentration. Bundles were counted as two filaments. Error bars show S.D. from at least 60 beads for each condition from three independent experiments. Coincidental filament overlap with control, BSA coated microsphere (⧫) was negligible. (C) Stacked bar chart showing average number of filaments (light gray) or bundles (dark gray) captured by WCA-coated microspheres, with indicated Mg²⁺, Lys-Lys²⁺, fascin concentrations. The proportion of captured bundles increased with increasing polycation or fascin.

https://doi.org/10.1371/journal.pone.0031385.g007

Among all three bundling factors, 1 µM fascin provided the highest number of captured barbed ends (27±5 per bead), a 10-fold increase over the number of barbed ends captured in 0.03 mM free Mg²⁺ (2.7±1.2 per bead). We note that while tail speeds in 0.03 mM Mg² decreased at higher fascin concentrations (Figure 4H), fascin addition in 0.03 mM Mg²⁺ monotonically both increased actin filament capture (Figure 7B) and prevented comet tail detachment (Figure 6M–O). Thus, the decrease in minimal Mg²⁺ motility rates in higher fascin shown in Figure 4 was likely due to fascin-induced differences in filament geometry or elongation within the comet tail and not due to failure of barbed end attachments to the bead surface at higher fascin levels.

Increased barbed end capture by increased Mg²⁺ was not due to an increase in the number of filaments available for capture as the Mg²⁺ range used did not affect actin polymerization (Figure S4A), nor was increased capture due solely to increased ionic strength as additional KCl only minimally increased the number of captured ends (Figure 7B). Further analysis of captured barbed ends (Figure 7C) showed that the number of single filaments captured did not vary substantially across all conditions. Beads captured from 2 to 1 individual filaments with increasing concentrations of Mg²⁺ or Lys-Lys²⁺ and from 2.5 to 0.7 individual filaments with increasing fascin. Rather, the increase in captured barbed ends resulted from a substantial increase in the number of captured bundles. Thus, capture of barbed ends by tethered WCA domains was largely due to filament bundling and WCA binding did not require Arp2/3 or CP.

Protein expression, purification, and fluorescent labeling

Actin was purified from rabbit skeletal muscle acetone powder through one round of polymerization, depolymerization, and gel filtration [97]. Actin was labeled with pyrenyl iodoacetamide (Invitrogen, Carlsbad, CA) [98] or with Oregon green 488 iodoacetamide (Invitrogen) as previously described [49]. Before use, labeled and unlabeled actins were dialyzed overnight against fresh buffer G (2 mM Tris-Cl pH 8, 0.2 mM ATP, 1 mM NaN₃, 0.1 mM CaCl₂, 0.5 mM dithiothreitol, DTT) and centrifuged at 100,000 g for 2 hr at 4°C. Arp2/3 complex was purified from bovine thymus as described [99]. Recombinant mouse capping protein was expressed in E. Coli and purified as described [100]. Human N-WASP-WCA was expressed as a glutathione S-transferase (GST) fusion proteins in E. Coli and purified on a Glutathione Agarose affinity column (Thermo Scientific, Rockford, IL) followed by anion exchange chromatography on a Source Q (GE Healthcare, Piscataway, NJ) column [33]. Recombinant human profilin I was expressed in E. Coli and purified by poly-L-proline affinity chromatography as described [101]. Recombinant human fascin I was expressed in E. Coli and purified and cleaved as described [102]. Actin and labeled actins were stored for 1 month at 4°C. All other proteins were flash frozen in liquid nitrogen and stored at −80°C.

Nanofiber and bead preparation

Nanofibers (200 nm nominal diameter, Johns Mansville, Denver, CO) were separated in chloroform with a Dounce homogenizer, centrifuged at 3750 rpm for 15 min in a clinical centrifuge, and the chloroform was evaporated in a fume hood. Nanofibers were washed once with deionized water by centrifugation and sonicated for 1 hour in 1 M KOH in a bath sonicator to remove contaminants. Nanofibers were washed briefly in deionized water, resuspended in 1 M HCl, sonicated for 1 hour, and incubated overnight in HCl. Cleaned nanofibers were subsequently pelleted by centrifugation and sonicated for 30 minutes each in in ddH₂O, 1 mM ethylene diamine tetraacetic acid (EDTA), 70% ethanol, and absolute ethanol to dry, with pelleting between each step. Cleaned nanofibers were stored in glass containers in absolute ethanol for up to three months.

Carboylated polystyrene 4.5 µm diameter microsphere (Polysciences, Warrington, PA) or glass nanofiber were coated respectively with 8.5 uM or 10 uM GST tagged WCA by incubation for 1 h at room temperature. Particles were pelleted by low speed centrifugation and resuspended in storage buffer (10 mM HEPES pH 7.8, 0.1 M KCl, 1 mM MgCl₂, 1 mM ATP, 0.1 mM CaCl₂, 0.01% NaN₃) containing 1 mg/ml bovine serum albumin (BSA, Sigma-Aldrich, St. Louis, MO) to block subsequent nonspecific binding. Particles were stored at 4°C for up to 1 week.

Reconstitution of nanofiber and bead motility under TIRF microscopy

Glass slides and coverslips were cleaned, and flow cells constructed as previously described [49]. For buffer exchange, flow cells were coated with 300 nM n-ethylmaleimide inactivated myosin II for 2 minutes. For all experiments, flow cells were coated with 1% BSA for 5–7 min as described [49]. After blocking, 16 µl of reaction mixture was wicked through the chamber and the chamber was either sealed with warm VALAP (1∶1∶1 vaseline/lanolin/paraffin) or left unsealed for subsequent buffer exchange. For bead motility assays, glass slides and coverslips were blocked overnight in 1% BSA at 4°C and dried in air before use. We placed 16 µl of reaction mixture on a BSA coated slide, covered with a BSA coated coverslip, and sealed the chamber with VALAP.

Labeled and unlabeled Ca-ATP actin were diluted to the desired labeled fraction, mixed 9∶1 with 10× magnesium exchange buffer (10× ME: 10 mM ethylene glycol tetraacetic acid, EGTA, 1 mM MgCl₂) and incubated on ice for 2 minutes to form 4× final concentrations of Mg-ATP actin. We placed 8 µl of Mg-ATP actin at the bottom of a 1.5 ml Eppendorf tube and added 7 µl of motility protein mixtures and 1 µl of coated nanofibers or beads on the side of the tube. Arp2/3 and CP were diluted in nanoparticle storage buffer (10 mM HEPES pH 7.8, 0.1 M KCl, 1 mM MgCl₂, 1 mM ATP, 0.1 mM CaCl₂, 0.01% NaN₃). We washed both drops together with 16 µl 2× TIRF buffer (2×: 100 mM KCl, 2 mM MgCl₂, 2 mM EGTA, 20 mM imidazole, pH 7.0, 200 mM DTT, 0.4 mM ATP, 30 mM glucose, 0.5% 1500 cP methylcellulose, 40 µg/ml catalase, 200 ug/ml glucose oxidase) and placed the reaction mixture in either a flow cell or slide-coverslip as described above.

Image acquisition and processing

Actin fluorescence was observed with a 60×1.49 NA TIRF objective on an Olympus IX2 inverted microscope. Images were captured with a Retiga EXi cooled CCD camera (QImaging) using SlideBook image acquisition software (Intelligent Imaging Innovations, Inc). All subsequent image-processing steps were performed in ImageJ, available at http://rsbweb.nih.gov/ij. TIRF microscopy images were gamma corrected using a value of between 0.5 and 0.8. An unsharp-mask filter was applied with a radius of 1 to 1.3 pixels and a 60% mask weight. epi-fluorescence microscopy images were unprocessed. For differential interference contrast (DIC) images, each image was divided by an averaged background image and contrast was adjusted to locate the nanofiber. Images were rotated and cropped for publication.

Bundle formation during nanofiber motility

To quantify the frequency of bundling, loop, and buckle formation (Table 2), we counted the number of nanofibers with obvious filament bundles and divided by the total number of nanofibers observed. For nanofibers with bundles, we classified those bundles as looped, buckled, or neither. Looped or buckled bundles exhibited significant curvature and had one end terminating at the nanofiber surface and the other end embedded in the comet tail. Looped bundles were longer and had one significant curve. Buckled bundles were shorter and had two or more curved sections.

Bead motility in low magnesium

For low Mg²⁺ motility buffers, we made the following changes: (1) For each experiment, we diluted fresh Ca-ATP actin into buffer G with no added Ca²⁺. (2) We mixed actin 9∶1 with 10× low magnesium exchange buffer (10× lowME: 2 mM EGTA, 0.2 mM MgCl₂) for 2 minutes to form 4× final concentration of actin. (3) We diluted all proteins in buffer G with no added Ca²⁺, or Mg²⁺, or EGTA. (4) We reduced MgCl₂ in 2× TIRF buffer from 2 mM to 0.2 mM to form 2× low-ME TIRF buffer. The final total concentration of Mg²⁺, Ca²⁺, EGTA, and ATP were 0.105 mM, 5 µM, 1.05 mM, and 0.2 mM, respectively. Free Mg²⁺ and Ca²⁺ concentrations were calculated using MaxChelator software [64], [65] available from maxchelator.stanford.edu.

Comet tail detachment

To quantify detachment of comet tails or actin shells from beads, we placed GST-WCA coated microspheres in low Mg²⁺ motility buffers as above and counted number of beads with comet tail or actin shell under epi-fluorescence microscopy every 30 minutes. We divided this count by the total number of beads observed per experiment. Actin shells were defined as a circumferential increase in actin polymerization around the bead surface. Shell detachment was scored as an absence of actin fluorescence in a ∼1 µm zone around the bead circumference. Detachment was often accompanied by a nearby empty shell or nascent comet tail. Average and S.D. of for each condition were calculated from three independent experiments. At least 50 beads were scored for each experiment.

Filament bundling in low and high magnesium

Labeled and unlabeled Mg-ATP-actin were diluted in buffer G without added calcium as above, mixed with low Mg²⁺ motility buffer supplemented with MgCl₂, added to chambers pre-blocked with 1% BSA, and filament growth was observed by TIRF microscopy. To quantify bundling, a 47 µm×35 µm subfield at the center of each video was chosen and the total numbers of individual filaments and the number of filaments converging into bundles were scored. Bundles were scored as filaments overlapping for two or more frames. The fraction of bundled filaments was scored from three independent experiments for each condition.

Filaments binding to microspheres

Coverslips were coated with 1 mg/ml poly-L-lysine (30 to 70 kDa, Sigma-Aldrich) for 2 minutes, rinsed 3× with deionized water, and dried in air. We polymerized 8 µM (30% Oregon green labeled) Mg-ATP actin in buffer F (buffer G with 10 mM Imidazole pH 7, 50 mM KCl, 0.105 mM MgCl₂, 1 mM EGTA) for 10 minutes at room temperature. Actin seeds were vortexed at maximum speed for 1 minute to break filaments. To prepared seeds, we added 4.5 µm diameter microspheres coated with 8.5 µM GST-WCA, 1 µM Oregon green labeled Mg-ATP actin monomers, and 1 mg/ml BSA. Mg²⁺, Lys-Lys²⁺, fascin or K⁺ were also added at this step. The reaction was incubated for 10 minutes at room temperature, beads and bound filaments and bundles were pelleted by centrifugation at 5000 rpm for 1 min, and the supernatant was removed. The pellet containing beads and bound filaments was gently diluted 16-fold in buffer F, mixed 1∶1 with 2× TIRF buffer in the absence of methylcellulose, and 16 µl was added to a poly-L-lysine coated coverslip. Filaments/bundles and beads were observed, respectively, by TIRF or DIC microscopy.

Bulk pyrene-actin spectroscopy assays

For actin and profilin-actin experiments, we diluted labeled and unlabeled Ca-ATP actin to 30% labeled fraction with or without added profilin, mixed 9∶1 with 10× low-Mg exchange buffer, and incubated on ice for 2 minutes to form 2× final concentrations of Mg-ATP actin in the lower row (preparatory wells) of a 96 well half area flat bottom plate (Corning). Various concentrations of MgCl₂ were placed in the upper row (reaction wells) of the same plate along with 1.6 µl of 100× antifoam (100×: 0.005% antifoam-204, Sigma-Aldrich), 2× initial concentration of low-Mg KMEI (10×: 500 mM KCl, 1 mM MgCl₂, 10 mM EGTA, 100 mM Imidazole, pH 7), and buffer G without added Ca²⁺. We started the reaction by transferring 80 µl of actin mixture from the lower preparatory row to the upper reaction row containing 80 µl in each well for a 160 µl total reaction. The reaction was gently mixed with a 12-channel pipette, and pyrene-actin fluorescence was measured in a Spectra MAX Gemini XPS fluorescent plate reader (Molecular Devices, Sunnyvale, CA) with excitation and emission wavelengths of 364 nm and 407 nm, respectively. Arp2/3 nucleation experiments were similarly performed with Arp2/3 and GST-WCA added to the reaction well prior to the reaction start. For capping protein activity assays, we added CP to the reaction well followed by short (vortexed) unlabeled actin filament seeds prior to the reaction start as previously described [103].